- Top of page
Objective To test whether a cognitive–behaviour therapy intervention program reduces the prevalence of depression during the first postnatal year in mothers of very preterm babies.
Design Prospective, single blind, randomised, controlled study.
Setting Perinatal centre in Western Australia.
Participants One hundred and ninety-nine out of 673 English-speaking mothers of infants admitted to the neonatal unit.
Intervention A six-session cognitive–behaviour therapy intervention program provided by a research midwife between weeks two and six after birth. Women in the control group received standard care.
Main outcome measures Depression and anxiety disorders occurring in the first year assessed by a clinical psychologist at structured interview using the Schedule for Affective Disorders and Schizophrenia (SADS) at 2 weeks, 2, 6 and 12 months.
Results One hundred and one women were randomised to the intervention group and 98 to the control group. Fifty-four mothers (27%) in the trial were diagnosed with minor or major depression in the 12 months following very preterm delivery, 29 (29%) in the intervention group and 25 (26%) in the control group (relative risk 1.1 [95% CI 0.80–1.5]). There were no differences in the time of onset or the duration of the episodes of depression between the groups. Overall, 74 mothers (37%) of the 199 met criteria for a diagnosis of psychological morbidity during the first year.
Conclusions Our intervention program did not alter the prevalence of depression in these mothers. Rates of depression and stress reactions are high in these mothers.
- Top of page
The postnatal period is a time of increased psychological vulnerability in women with the most common complication being postnatal depression.1 Postnatal depression is reported to occur in 10–16% of all childbearing women.2
Research has identified numerous psychological and social factors that are associated with an increased prevalence of postnatal depression including a history of depression, marital problems, poor social support and infant problems.2,3 Birth is generally seen as a positive experience especially when parents are able to plan ahead. The early birth, at a very preterm gestation, can alter the usual response of the mother, create uncertainty about the infant's survival, guilt and disappointment about the birth and severe disruption to normal routines.4
Initially, mothers' responses following very preterm birth were viewed as an acute emotional reaction to a specific stressful event.5 The mother is often ill herself or recovering from an emergency caesarean section and this in itself is associated with increased psychological sequelae including postnatal depression.6 A high prevalence of depressed and anxious mood during the first year after very preterm delivery has been reported in small samples of mothers.7,8
Attempts to prevent depression in women identified as being at high risk of postnatal psychological disorder have used a variety of interventions including drug therapy.9–16 The results have not been positive. None have been specifically directed at mothers who delivered an infant at a very preterm gestation although there are two studies that have focussed on changing the short term impact of very preterm birth.17,18
Cognitive–behaviour therapy is recognised as an effective treatment for depression and has been shown to be effective in subsyndromal depression especially in those who are stressed.19,20
We report a randomised, single blind, controlled trial conducted to assess whether a cognitive–behaviour therapy group intervention program could reduce the prevalence of depression in mothers of very preterm infants. We also wished to ascertain the prevalence of depression and anxiety disorders in the mother during the first 12 months postpartum.
- Top of page
All mothers who had a very preterm (<33 weeks) or very low birthweight (<1500 g) infant admitted to the neonatal intensive care unit at King Edward Memorial Hospital for Women from 1st March 1993 until 30th June 1995 were eligible (Table 1). Approximately 95% of all very preterm infants born in Western Australia are cared for at this hospital.
Table 1. Inclusion and exclusion criteria.
|Admission of very preterm or very low birthweight infant|
|Lived in or near Perth (capital city of Western Australia)|
|History of psychotic disorder|
|Current treatment for unipolar or bipolar depression|
|Significant substance abuse|
|Infant unlikely to survive first week|
|Infant requiring early transfer to Children's Hospital for surgery|
|Assessment measures culturally inappropriate|
|Maternal age <17 years|
A research midwife approached mothers in the first week after delivery. Once written informed consent was obtained demographic, antenatal, perinatal, past medical and family history questionnaires were completed. A clinical psychologist interviewed each woman during the second postpartum week, using a structured psychological interview (Schedule for Affective Disorders and Schizophrenia [SADS]).21 Diagnoses were based on criteria published in the fourth edition of the Statistical Manual of Mental Disorders (DSM-IV).22 The research psychologists were trained in the use of this structured interview during a pilot study. Women not depressed at this initial postpartum assessment were then immediately individually randomised to either the intervention group or the control group.
Random assignment was by computer generated cards. A selection of these was presented to the mother in opaque sealed envelopes. Neither the mother nor the research midwife was aware of the allocation before opening the envelope. Randomisation was stratified on the basis of gestational age at delivery, <28 weeks or not, and parity.
Women assigned to the intervention group were offered a brief individual debriefing discussion related to their pregnancy experience. The research midwife then facilitated six weekly group sessions, each of 2 hours duration. A cognitive–behaviour therapy model provided the basis for these sessions but they also included an educational component and advice on how to cope with the practical problems of the first few weeks and months. The sessions dealt with the following issues: adjustment to parenthood with a preterm infant; explaining postnatal depression; coping with emotional and physical changes; identifying and altering negative patterns of thinking; daily and weekly activity planning and developing self-nurturing strategies; communication issues in relationships and setting short term and long term goals. The research midwives were trained in these techniques by one of the chief investigators (SP) and she continued supervision during the study.
Women assigned to the control group received the standard care for mothers of preterm infants in our hospital. This included social work contact for all mothers, regular biweekly parent education group sessions and a developmental physiotherapy playgroup during the first year.
All women were assessed at 2, 6 and 12 months. At each assessment, self-report depression questionnaires (Edinburgh Postnatal Depression Scale, Beck Depression Inventory, General Health Questionnaire—Scaled version) were completed and a clinical psychologist interviewed each mother using the structured interview schedule. Higher scores on the questionnaires are indicative of more distress. Women in the intervention group completed an evaluation questionnaire at the end of the program.
The research midwife was the only member of the team to know an individual mother's allocation. The clinical psychologist was blind to the mother's intervention status and the results of the self-report questionnaires. The effectiveness of this was reviewed at the weekly team research meetings and there were no overt breaches reported.
The primary outcome was the diagnosis of major or minor depression, using DSM-IV criteria, in the 12 months following the very preterm delivery. Women who met criteria for minor depression had significantly depressed mood with functional impairment.
We hypothesised that the prevalence of depression would be 35%. Based on previous work on the effectiveness of counselling in women with postnatal depression,23 a power calculation before the study showed that we would need 98 mothers in each group to detect a difference in outcome of 50% at 80% power using a χ2 test with continuity correction and an α= 0.05. The baseline rate was based on that measured using screening questionnaires in a cohort of mothers taking part in another study of very preterm infants and the targeted reduction rate was close to, but slightly higher than, the prevalence in mothers of term infants in our community.24
The data were analysed using univariate and multivariate statistical techniques within the Statistical Analysis System. Analyses were two tailed with the significance level set at 0.05 for all tests and were based on intention to treat.
Numeric variables are presented as median and interquartile range and compared using the Wilcoxon rank sum test. Group testing of categorical data is presented as n (%) and compared with Fisher's exact test or χ2 test. Relative risks were calculated using logit estimates with the corresponding confidence limits. The homogeneity of relative risks over strata was checked using the Breslow–Day test. Ethical approval was obtained from the Institutional Ethics Committee.
Our null hypothesis was that there would be no difference in the incidence of depression between women who participated in the intervention program and those who received routine postpartum care.
- Top of page
Six hundred and seventy-three mothers had 782 infants admitted to the neonatal intensive care unit during the study. Two hundred and thirty-seven (35%) mothers were ineligible (Fig. 1).
Of the 436 women eligible, 222 (51%) agreed to be randomised while 61 (14%) refused. The remainder (153; 35%) agreed to complete self-report assessment questionnaires during the first year, but did not wish to participate in the randomised controlled trial. Mothers in the randomised trial were significantly different from mothers who agreed to complete self-report questionnaires being older, more likely to be primiparous, having a positive psychological history and delivering at a slightly earlier gestational age. During the first year, there were only small and inconsistent non-significant differences in the scores on the self-report questionnaires between mothers in the study and those answering the screening questionnaires.
Twenty-three of the mothers who agreed to participate were depressed at the initial interview and were not randomised. Thus, 101 mothers were randomised to the intervention group and 98 to the control group.
Women randomised to the intervention group or the control group showed no major differences except more mothers in the control group had previously had a preterm infant (Table 2). There were no differences in scores on depression screening questionnaires between the two groups at two weeks postpartum. Eighty percent of the women attended at least three of the six intervention sessions and 60% attended all sessions.
Table 2. Social, clinical and psychological factors in women in the control and intervention groups. Values are expressed as n (%), mean [SD] or median (interquartile range).
|Maternal age||29.5 (26–34)||29 (25–32)||0.226|
|Married/de facto||88 (90)||91 (89)||1.00|
|Social class||2.56 [0.9]||2.53 [0.89]||0.813|
|Maternal smoking||19 (19)||16 (16)||0.578|
|Maternal education|| || ||0.852|
|School only||41 (42)||44 (43)|| |
|Trade certificate/diploma||41 (42)||42 (41)|| |
|Tertiary||16 (16)||16 (16)|| |
|Primigravid||41 (42)||51 (50)||0.260|
|Primiparous||61 (62)||67 (66)||0.660|
|Previous preterm infant||15 (15)||6 (6)||0.038|
|Planned pregnancy||65 (66)||73 (72)||0.448|
|Assisted pregnancy||15 (15)||22 (22)||0.279|
|Exposed to labour||51 (52)||62 (61)||0.254|
|Caesarean section||57 (58)||58 (57)||0.887|
|Gestational age delivery||30 (27–31)||29 (28–31)||0.648|
|Twins||16 (16)||18 (18)||0.827|
|Previous depression||26 (26)||33 (33)||0.471|
|Antenatal depression||9 (9)||12 (12)||0.058|
|2 week EPDS||8 (4–12)||8 (4–11)||0.499|
|GHQ—Likert||21 (15–30)||22 (14–31)||0.749|
Diagnoses were available for 176 mothers to 12 months and 192 mothers to 6 months postpartum. Eleven mothers withdrew during the study, nine in the intervention arm and two controls. The follow up rate was 99% at 2 months, 96% at 6 months and 88% at 12 months (Fig. 1). Fifty-four mothers (27%) in the trial were diagnosed with major or minor depression in the 12 months following very preterm delivery, 29 (29%) in the intervention group and 25 (26%) in the control group (P= 0.563; relative risk 1.13 [95% CI 0.71–1.78]). There were no differences between the groups in depression diagnoses or in depression screening questionnaires at any time period (Table 3). There was no effect of the number of intervention sessions attended and the development of postnatal depression (P= 0.962).
Table 3. Number of women depressed and screening questionnaires scores at each assessment time. Values are presented as n or median (interquartile range). Median (interquartile range) shown for questionnaires, higher scores reflect more problems.
| EPDS||5 (2–8)||6 (3–9)||0.345|
| BDI||4 (1–6)||5 (2–8)||0.089|
| GHQ||13 (10–20)||14 (10–21)||0.430|
| EPDS||4 (2–8)||4 (3–8)||0.215|
| BDI||4 (1–9)||5 (2–9)||0.243|
| GHQ||14 (9–21)||14 (10–20)||0.895|
| EPDS||5 (1–9)||4 (1–8)||0.591|
| BDI||4 (1–7)||3 (2–8)||0.609|
| GHQ||14 (11–21)||14 (10–21)||0.676|
A history of depression prior to this pregnancy or an episode of depression in the antenatal period of this pregnancy was associated with a higher level of depression following this birth. This was 43% (13/30) in the control group and 38% (16/42) in the intervention group. Controlling for these confounders did not change the outcome, adjusted relative risk 1.02 (95% CI 0.87–1.20). Models controlling for other confounders also did not alter the results. There were no differences in the time of onset or the duration of the episodes of depression between the groups. The median (interquartile range) time of onset was 18 (6–31) weeks in the control group and 14 (9–30) weeks in the intervention group with duration of 7 (4–20) weeks and 7 (4–19) weeks, respectively.
Overall, 74 mothers (37%) of the 199 met rigorous criteria for a diagnosis of psychological morbidity during the first year (Table 4). These included, in addition to major and minor depression, acute stress disorder, post-traumatic stress disorder and adjustment reactions with anxiety. When the 23 mothers who agreed to participate in the randomised trial but were diagnosed with depression at two weeks, and thus not randomised, are included then 44% of women had a diagnosis of psychological morbidity.
Table 4. Total psychological morbidity in women who agreed to participate in trial.
| ||Control||Intervention||Depressed at initial assessment||Total||%|
|N||98||101||23|| || |
Mothers rated the intervention program as helpful and appropriate as did the two experienced midwives who facilitated the sessions.
- Top of page
This is the first randomised trial reported in mothers of very preterm infants that solely focussed on the prevention of depression with follow up throughout the entire first year after delivery. Significant psychological morbidity, particularly depressive and anxiety disorders, is extremely common in mothers of these infants in the year following childbirth. Thirty-five percent (77/222) of the mothers enrolled in this trial were diagnosed with depressive disorders, an incidence clearly higher than the 18% we have reported in mothers of term infants.24 Our cognitive–behaviour therapy based intervention had no effect on the prevalence of depression during the study period and was not associated with any deleterious effects.
Only half of the women who were eligible for inclusion participated in the randomised trial. While they had some significant differences from those not involved, the scores on the self-report screening questionnaires would indicate that all these women experienced high levels of distress during the first year and that our results could be generalised to our total population. Other researchers have commented on difficulties enrolling perinatal women in randomised trials and that those who participate may not be representative of the target group in that those refusing randomisation frequently have significantly higher rates of morbidity.25 This does not seem to have happened in our study, which enrolled a high proportion of women with a history of depression including depression during this pregnancy.
Short term psychological benefits in mothers of very preterm infants have been reported using a parent buddy program and an individualised family-based intervention.17,18 Both studies however enrolled small numbers of mothers, used only self-report screening questionnaires to assess mood and followed mothers for only a short period after birth.
Recent studies have reported on a variety of interventions in other high risk groups of mothers.9–16 These mothers were selected on the basis of positive antenatal screening for risk factors associated with postnatal depression,10,11,14 a previous history of depression16 or the presence of perinatal risk factors associated with postnatal depression.9,12,13,15 Four of these studies used structured psychiatric interviews to assess outcome10–13,16 but with one exception9 followed mothers only for the first four months after the birth. Only one study reported benefits from the intervention in reduced scores on the self-report screening questionnaire in the first six weeks.12,13 The cognitive–behaviour therapy based group intervention program we tested did not reduce the prevalence of depression as assessed by structured interview nor show any difference on scores on the self-report screening questionnaires at any of our assessment time periods.
There are differences in how screening questionnaires and structured psychiatric interviews identify morbidity both in the time period assessed and the rigour of the diagnosis. Screening questionnaires are appropriate for large scale surveys but are less useful for rigorously testing the effects of an intervention program in a randomised controlled trial. Calculating study power on changes on a continuous variable such as a score on a screening questionnaire demand much smaller sample sizes although the clinical significance of a small or even moderate difference on these scores is uncertain.
The duration of the follow up period is also important, as short term effects may not persist.26 The extended length of follow up assessment used in our study meant that cases of depression with onset later in the first year were accurately detected and we could accurately assess the duration of the initial episode. This is important in this clinical group as the very preterm infant is frequently not discharged until many months of age and it may be that providing care for an infant with continuing health problems is associated with later onset in many cases. The median time of onset in mothers in our study was 16 weeks after delivery significantly longer than the 6 weeks we have reported in mothers of term infants in our community.24 The studies reported in other high risk groups of mothers did not comment on the time of onset or duration of the depression in their mothers. Their short duration of follow up would have prevented this and it is unclear if other high risk mothers show a similar delayed time of onset as we report. Our extended period of follow up in this study and the rigorousness of our assessments ensured also that we would find any deleterious effects of our intervention program. We did not find any evidence of this.
Why was our intervention not effective? Our intervention program was modelled on a depression treatment program27 that had been modified for use in our institution with mothers with postnatal depression. Treatment based on similar principles is known to be effective in postnatal depression.26 This may not however be an appropriate approach in attempting to prevent the onset of depression as there are concerns regarding the timing, content and facilitation of postnatal preventive interventions.10 Our intervention sessions were held when most mothers were still preoccupied with worry about their infant or caring for them in the neonatal intensive care unit. This prevented mothers from attending some sessions even though post hoc analyses showed no effect of either the number of sessions attended, or a subjective assessment of the mother's commitment to the program on the outcome. It is therefore difficult to ensure optimal timing for presentation of the prevention strategies and given the late onset of depression in many mothers our intervention may have been more effective if it had contained follow up sessions later in the first year. However, there were no differences between the groups at any of our assessment periods. The mothers rated the content of our intervention helpful and appropriate, yet it had no effect on their outcomes. Preventive programs may need to identify and strengthen those characteristics that increase resilience and ability to adapt to change instead of targeting decreasing negative behaviours and/or cognitions.28 The facilitators we used were experienced midwives, used to working in a tertiary centre, who had been trained in facilitating this program and were supervised by a clinical psychologist (SP) during the study. A pilot study before the trial started had suggested that the timing and content of the program and the use of midwives were appropriate. In addition, during the study the mothers completed homework after each session to help us ensure that the program was being understood. All mothers had structured psychological interviews at 2, 6 and 12 months and these in themselves may have been therapeutic and eliminated any effect of our intervention. Yet despite this the prevalence of depression and other psychological morbidity was extremely high.
Parents of preterm infants are extremely distressed after the birth of their infants and the mother may blame herself for this early birth.4,5,29 Concerns for their infant centre on survival and long term care and are frequently still present at discharge.5 The high rate of depression we have found is thus not surprising and is consistent with other studies.7 What is new in this study is the very high rate of other psychological morbidity in addition to depression, such that almost one in two mothers met rigorous criteria for a diagnosis of a psychological disorder during the first year. Many of these cases were instances of co-morbid depression and anxiety disorders that might be expected after such a catastrophic life event. No previous study has applied such an intensive assessment schedule nor documented the high rate of morbidity.
Our current follow up program for very preterm infants has not identified this high prevalence of mothers with significant psychological difficulties in the first year postpartum. The mother's psychological state affects her capacity to access support and her ability to relate to her partner and to appropriately relate to her infant.30 A high prevalence of behaviour disorders has been reported in low birthweight infants and maternal depression has been identified as a major factor in these disorders.31,32 The high level of psychological morbidity during the first year of the child's life, recorded in our study, may thus have major consequences for the child in later years.
We conclude that a cognitive–behaviour based intervention program, based on treatment approaches, does not influence the high rate of psychological morbidity in mothers of very preterm or very low birthweight infants. We have incorporated some of the elements of our program into our regular parent evening education sessions and have added mental health professionals to our clinical team. We have done this to establish some degree of therapeutic alliance between mental health professionals and the mother as this has been reported as being beneficial in helping mothers accept treatment.12 We are considering introducing maternal self-report screening questionnaires as part of routine follow up of these infants to identify early those mothers with depression or anxiety. Thus, we favour early detection and treatment of established cases until further research identifies effective and feasible interventions for prevention.